Compute Library
 21.02
NEQLSTMLayerNormalizationKernel Class Reference

Neon kernel to perform layer normalization. More...

#include <NEQLSTMLayerNormalizationKernel.h>

Collaboration diagram for NEQLSTMLayerNormalizationKernel:
[legend]

Public Member Functions

const char * name () const override
 Name of the kernel. More...
 
 NEQLSTMLayerNormalizationKernel ()=default
 Default constructor. More...
 
 NEQLSTMLayerNormalizationKernel (const NEQLSTMLayerNormalizationKernel &)=delete
 Prevent instances of this class from being copied (As this class contains pointers) More...
 
NEQLSTMLayerNormalizationKerneloperator= (const NEQLSTMLayerNormalizationKernel &)=delete
 Prevent instances of this class from being copied (As this class contains pointers) More...
 
 NEQLSTMLayerNormalizationKernel (NEQLSTMLayerNormalizationKernel &&)=default
 Default Move Constructor. More...
 
NEQLSTMLayerNormalizationKerneloperator= (NEQLSTMLayerNormalizationKernel &&)=default
 Default move assignment operator. More...
 
 ~NEQLSTMLayerNormalizationKernel ()=default
 Default destructor. More...
 
void configure (const ITensor *input, ITensor *output, const ITensor *weight, const ITensor *bias)
 Set the input and output tensors. More...
 
void run (const Window &window, const ThreadInfo &info) override
 Execute the kernel on the passed window. More...
 
- Public Member Functions inherited from ICPPKernel
virtual ~ICPPKernel ()=default
 Default destructor. More...
 
virtual void run_nd (const Window &window, const ThreadInfo &info, const Window &thread_locator)
 legacy compatibility layer for implemantions which do not support thread_locator In these cases we simply narrow the interface down the legacy version More...
 
virtual void run_op (ITensorPack &tensors, const Window &window, const ThreadInfo &info)
 Execute the kernel on the passed window. More...
 
- Public Member Functions inherited from IKernel
 IKernel ()
 Constructor. More...
 
virtual ~IKernel ()=default
 Destructor. More...
 
virtual bool is_parallelisable () const
 Indicates whether or not the kernel is parallelisable. More...
 
virtual BorderSize border_size () const
 The size of the border for that kernel. More...
 
const Windowwindow () const
 The maximum window the kernel can be executed on. More...
 

Static Public Member Functions

static Status validate (const ITensorInfo *input, const ITensorInfo *output, const ITensorInfo *weight, const ITensorInfo *bias)
 Static function to check if given info will lead to a valid configuration of NEQLSTMLayerNormalizationKernel. More...
 

Detailed Description

Neon kernel to perform layer normalization.

Definition at line 35 of file NEQLSTMLayerNormalizationKernel.h.

Constructor & Destructor Documentation

◆ NEQLSTMLayerNormalizationKernel() [1/3]

Default constructor.

Referenced by NEQLSTMLayerNormalizationKernel::name().

◆ NEQLSTMLayerNormalizationKernel() [2/3]

Prevent instances of this class from being copied (As this class contains pointers)

◆ NEQLSTMLayerNormalizationKernel() [3/3]

Default Move Constructor.

◆ ~NEQLSTMLayerNormalizationKernel()

Default destructor.

Referenced by NEQLSTMLayerNormalizationKernel::name().

Member Function Documentation

◆ configure()

void configure ( const ITensor input,
ITensor output,
const ITensor weight,
const ITensor bias 
)

Set the input and output tensors.

Parameters
[in]inputSource tensor. Data types supported: QSYMM16.
[out]outputDestination tensor. Data types supported: Same as input.
[in]weightWeight tensor. Data types supported: Same as input.
[in]biasBias tensor. Data types supported: S32

Definition at line 84 of file NEQLSTMLayerNormalizationKernel.cpp.

References ARM_COMPUTE_ERROR_ON, ARM_COMPUTE_ERROR_ON_NULLPTR, ARM_COMPUTE_ERROR_THROW_ON, ITensor::info(), arm_compute::test::validation::input, arm_compute::QSYMM16, and arm_compute::validate().

Referenced by NEQLSTMLayerNormalizationKernel::name().

85 {
86  ARM_COMPUTE_ERROR_ON_NULLPTR(input, weight, bias, output);
87  ARM_COMPUTE_ERROR_ON(input == output);
88  ARM_COMPUTE_ERROR_THROW_ON(validate(input->info(), output->info(), weight->info(), bias->info()));
89 
90  static const std::map<DataType, ComputeFuncType> fn_map =
91  {
92  { DataType::QSYMM16, std::mem_fn(&NEQLSTMLayerNormalizationKernel::compute_qsymm16) },
93  };
94 
95  _input = input;
96  _output = output;
97  _weight = weight;
98  _bias = bias;
99  _fn = fn_map.at(_input->info()->data_type());
100 
101  auto_init_if_empty(*_output->info(), *_input->info());
102  _output->info()->set_quantization_info(compute_output_qinfo());
103 
104  const UniformQuantizationInfo wq_info = _weight->info()->quantization_info().uniform();
105  const Status s = quantization::calculate_quantized_multiplier(wq_info.scale, &_output_multiplier, &_output_shift);
106  _output_shift *= -1;
107 
108  if(!bool(s))
109  {
110  _output_multiplier = 0;
111  _output_shift = 0;
112  }
113 
114  Window win = configure_window(output);
115  INEKernel::configure(win);
116 }
static Status validate(const ITensorInfo *input, const ITensorInfo *output, const ITensorInfo *weight, const ITensorInfo *bias)
Static function to check if given info will lead to a valid configuration of NEQLSTMLayerNormalizatio...
quantized, symmetric fixed-point 16-bit number
virtual DataType data_type() const =0
Data type used for each element of the tensor.
#define ARM_COMPUTE_ERROR_ON(cond)
If the condition is true then an error message is printed and an exception thrown.
Definition: Error.h:466
#define ARM_COMPUTE_ERROR_THROW_ON(status)
Definition: Error.h:455
Status calculate_quantized_multiplier(float multiplier, int32_t *quant_multiplier, int32_t *shift, bool ignore_epsilon=false)
Calculate quantized representation of multiplier.
UniformQuantizationInfo uniform() const
Return per layer quantization info.
bool auto_init_if_empty(ITensorInfo &info, const TensorShape &shape, int num_channels, DataType data_type, QuantizationInfo quantization_info=QuantizationInfo())
Auto initialize the tensor info (shape, number of channels and data type) if the current assignment i...
virtual ITensorInfo * info() const =0
Interface to be implemented by the child class to return the tensor&#39;s metadata.
virtual ITensorInfo & set_quantization_info(const QuantizationInfo &quantization_info)=0
Set the quantization settings (scale and offset) of the tensor.
virtual QuantizationInfo quantization_info() const =0
Get the quantization settings (scale and offset) of the tensor.
#define ARM_COMPUTE_ERROR_ON_NULLPTR(...)
Definition: Validate.h:161

◆ name()

◆ operator=() [1/2]

Prevent instances of this class from being copied (As this class contains pointers)

Referenced by NEQLSTMLayerNormalizationKernel::name().

◆ operator=() [2/2]

Default move assignment operator.

◆ run()

void run ( const Window window,
const ThreadInfo info 
)
overridevirtual

Execute the kernel on the passed window.

Warning
If is_parallelisable() returns false then the passed window must be equal to window()
Note
The window has to be a region within the window returned by the window() method
The width of the window has to be a multiple of num_elems_processed_per_iteration().
Parameters
[in]windowRegion on which to execute the kernel. (Must be a region of the window returned by window())
[in]infoInfo about executing thread and CPU.

Reimplemented from ICPPKernel.

Definition at line 166 of file NEQLSTMLayerNormalizationKernel.cpp.

References ARM_COMPUTE_ERROR_ON, ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW, ARM_COMPUTE_ERROR_ON_MSG, ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL, ARM_COMPUTE_UNUSED, arm_compute::quantization::multiply_by_quantized_multiplier(), arm_compute::multiply_by_quantized_multiplier_2row(), arm_compute::wrapper::vadd(), arm_compute::wrapper::vcombine(), arm_compute::wrapper::vdup_n(), arm_compute::wrapper::vgethigh(), arm_compute::wrapper::vgetlane(), arm_compute::wrapper::vgetlow(), arm_compute::wrapper::vloadq(), arm_compute::wrapper::vmovl(), arm_compute::wrapper::vmovn(), arm_compute::wrapper::vmul(), arm_compute::wrapper::vpaddl(), arm_compute::wrapper::vqmovn(), arm_compute::wrapper::vstore(), arm_compute::wrapper::vsub(), and IKernel::window().

Referenced by NEQLSTMLayerNormalizationKernel::name().

167 {
171  ARM_COMPUTE_ERROR_ON_MSG(!_fn, "internal function is not defined for computation");
172 
173  _fn(*this);
174 }
const Window & window() const
The maximum window the kernel can be executed on.
Definition: IKernel.cpp:28
#define ARM_COMPUTE_UNUSED(...)
To avoid unused variables warnings.
Definition: Error.h:152
#define ARM_COMPUTE_ERROR_ON_MSG(cond, msg)
Definition: Error.h:456
#define ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(k)
Definition: Validate.h:941
ScaleKernelInfo info(interpolation_policy, default_border_mode, PixelValue(), sampling_policy, false)
#define ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(f, s)
Definition: Validate.h:205

◆ validate()

Status validate ( const ITensorInfo input,
const ITensorInfo output,
const ITensorInfo weight,
const ITensorInfo bias 
)
static

Static function to check if given info will lead to a valid configuration of NEQLSTMLayerNormalizationKernel.

Parameters
[in]inputSource tensor info. Data types supported: QSYMM16.
[in]outputDestination tensor info. Data types supported: Same as input.
[in]weightWeight tensor info. Data types supported: Same as input.
[in]biasBias tensor info. Data types supported: S32
Returns
a status

Definition at line 140 of file NEQLSTMLayerNormalizationKernel.cpp.

References ARM_COMPUTE_ERROR_ON_NULLPTR, ARM_COMPUTE_RETURN_ERROR_ON, ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN, ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES, ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_SHAPES, ARM_COMPUTE_UNUSED, ITensorInfo::num_dimensions(), arm_compute::QSYMM16, arm_compute::S32, ITensorInfo::tensor_shape(), ITensorInfo::total_size(), and Dimensions< T >::x().

Referenced by NEQLSTMLayerNormalizationKernel::name().

141 {
142  ARM_COMPUTE_UNUSED(output, bias, weight, input);
143 
144  ARM_COMPUTE_ERROR_ON_NULLPTR(input, weight, bias, output);
145 
149 
150  ARM_COMPUTE_RETURN_ERROR_ON(input->num_dimensions() > max_input_dimension);
151  ARM_COMPUTE_RETURN_ERROR_ON(weight->num_dimensions() > max_weight_dimension);
152  ARM_COMPUTE_RETURN_ERROR_ON(bias->num_dimensions() > max_bias_dimension);
153 
154  ARM_COMPUTE_RETURN_ERROR_ON(input->tensor_shape().x() != weight->tensor_shape().x());
156 
157  if(output->total_size() != 0)
158  {
161  }
162 
163  return Status{};
164 }
quantized, symmetric fixed-point 16-bit number
#define ARM_COMPUTE_RETURN_ERROR_ON(cond)
If the condition is true, an error is returned.
Definition: Error.h:296
1 channel, 1 S32 per channel
#define ARM_COMPUTE_UNUSED(...)
To avoid unused variables warnings.
Definition: Error.h:152
#define ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_SHAPES(...)
Definition: Validate.h:443
#define ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(...)
Definition: Validate.h:545
#define ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(t, c,...)
Definition: Validate.h:792
#define ARM_COMPUTE_ERROR_ON_NULLPTR(...)
Definition: Validate.h:161

The documentation for this class was generated from the following files: